Image: from Alex Xplore X @AlexXplore
By Harrison Vane | Open Chronicle Technology & Defense Correspondent
AUSTIN, TX — In a characteristic blend of industrial ambition and science-fiction prophecy, Elon Musk has officially launched Terafab, a multi-billion-dollar semiconductor initiative designed to centralize chip production for his sprawling corporate empire. Standing before the defunct Seaholm Power Plant in Austin on March 21, Musk declared that the joint venture between Tesla, SpaceX, and the recently acquired xAI would become the most advanced chip fabrication facility in human history.
The project, estimated to cost between $20 billion and $25 billion, aims to solve a bottleneck Musk describes as an existential threat: the world’s inability to manufacture AI chips at the “galactic scale” his ventures require.
Bro… Giga Texas is already 3X the size of the Pentagon, over 2,500 acres, where its main building is over 10M sq ft, equivalent to about 100 football fields under one roof.
Now Elon is saying the Terafab will be 100M sq ft, that’s 10x larger than this!
pic.twitter.com/bbnYVcuCuh— Teslaconomics (@Teslaconomics) March 23, 2026
Everything Under One Roof
The Terafab will be located on the North Campus of Giga Texas, effectively turning the Austin site into a sovereign tech enclave. Musk’s primary critique of the current semiconductor industry is its fragmentation. Traditionally, a chip might be designed in California, etched in Taiwan, and packaged in Malaysia.
Terafab intends to collapse this supply chain into a single building. The facility will integrate:
-
Proprietary Lithography: Utilizing 2-nanometer nodes.
-
Logic and Memory Production: Custom-built for AI training and edge inference.
-
Advanced Packaging and Testing: Allowing for what Musk calls a “recursive improvement loop,” where a chip can be designed, printed, tested, and redesigned in a matter of days rather than months.
We're building TERAFAB to close the gap between today’s chip production & the future's demand – a future among the stars.
Join us → https://t.co/512DIlqNgY pic.twitter.com/ATr0e0pRDJ
— SpaceX (@SpaceX) March 22, 2026
Two Chips, Two Worlds
During the presentation, Musk outlined a production strategy focused on two distinct hardware families:
-
The Terrestrial Inference Chip (AI5/AI6): Optimized for “edge compute,” these processors will power the Tesla Cybercab, the Full Self-Driving (FSD) suite, and the Optimus humanoid robot. Musk predicts a future where Optimus robots outnumber humans, requiring hundreds of millions of chips annually.
-
The D3 Space-Hardened Processor: A high-power chip designed to survive the radiation and thermal extremes of orbit. These will power SpaceX’s ambitious “AI Sat Mini” constellation—a fleet of millions of satellites intended to act as an orbital data center.
The Terawatt Ambition
The name “Terafab” refers to the project’s goal of producing one terawatt of annual computing power. To put this in perspective, Musk noted that the current global AI compute output is roughly 20 gigawatts, meaning Terafab aims to exceed the entire world’s current capacity by a factor of 50.
“We either build the Terafab, or we don’t have the chips,” Musk told the cheering crowd. “And we need the chips to become a multi-planetary civilization.”
In a move that raised eyebrows among environmentalists and economists, Musk revealed that 80% of Terafab’s compute output is destined for space. He argued that solar irradiance is five times higher in orbit and the vacuum provides a more efficient heat sink, potentially making space-based AI cheaper than terrestrial data centers within three years.
Skeptics and the “Reality Gap”
Despite the spectacle, which included light beams shooting into the Austin night sky, industry veterans remain cautious. High-end chip fabrication is notoriously difficult; even giants like Intel and Samsung have struggled with 2nm yields. Furthermore, the project comes amidst a global helium shortage, a critical component in chip making, exacerbated by ongoing regional conflicts.
Critics also pointed to Musk’s history of “Elon time,” noting that while small-batch production of the AI5 chip is slated for late 2026, the full-scale Terafab facility could take a decade to reach the staggering volumes promised.
Regardless of the timeline, the message to the market was clear: Musk is no longer content being a customer of the silicon industry. He intends to own it.